LogoLMSpeed
Back to models

Model compare

Claude Sonnet 4.6 vs GPT-5.4 Mini

The readout for Claude Sonnet 4.6 and GPT-5.4 Mini, before the detailed comparison sheet.

Model A

Claude Sonnet 4.6

Anthropic

Contender
vs

Model B

GPT-5.4 Mini

OpenAI

Leading

Key Takeaways

Weighted outcome: GPT-5.4 Mini. Benchmark capability categories carry 80%, while price, API performance, and availability carry 20%.

Decision read

GPT-5.4 Mini

GPT-5.4 Mini has the higher weighted result; Model A / B score 34.3 to 65.7.

Evidence depth

137 data points

Includes 36 benchmark rows, 0 audit samples, and 10 provider examples.

Selection signal

Start with GPT-5.4 Mini

The charts below split 46 high-signal samples across speed, scores, and audit health.

Change comparison

Switch either side of this report to compare another model with the same LMSpeed data pipeline.

Select a different model to open a new comparison URL.

Comparison sheet

This report only uses LMSpeed data for Claude Sonnet 4.6 and GPT-5.4 Mini: pricing, speed aggregates, third-party benchmark scores, and shared provider samples.

Model compare
Claude Sonnet 4.6
GPT-5.4 Mini
Overall leaderContenderLeading
Weighted overall score34.3 pts65.7 pts
Benchmark category leads3 categories4 categories
Operational advantagesNo dataCheapest input price, Average speed, First-token latency, Free providers, Provider coverage
Context window1M tokens400K tokens
Max output128K tokens128K tokens
Modalities

Input

TextImageFile

Output

Text

Input

FileImageText

Output

Text
Features
Text inputImage inputFile inputText outputTool callingStructured outputsJSON modeReasoning
Text inputImage inputFile inputText outputTool callingStructured outputsJSON modeReasoning

The overall result weights benchmark capability categories at 80% and price, API speed/latency, and availability at 20%. Recent test volume does not affect the winner, and missing benchmark categories are excluded.

Model metadata

Model compare
Claude Sonnet 4.6
GPT-5.4 Mini
DeveloperAnthropicOpenAI
ReleasedFeb 2026Mar 2026
ParametersNo dataNo data
TokenizerClaudeGPT
Knowledge cutoffNo data2025-08-31
OpenRouter IDanthropic/claude-sonnet-4.6openai/gpt-5.4-mini
ReferencesNo dataNo data

When to choose each model

This report only uses LMSpeed data for Claude Sonnet 4.6 and GPT-5.4 Mini: pricing, speed aggregates, third-party benchmark scores, and shared provider samples.

Claude Sonnet 4.6

Claude Sonnet 4.6 leads these benchmark categories: Agents, Knowledge, Math.

GPT-5.4 Mini

GPT-5.4 Mini is stronger in benchmark categories (Coding, Reasoning, Multimodal, Instruction following) and operational dimensions (Cheapest input price, Average speed, First-token latency, Free providers, Provider coverage).

Benchmark score comparison

Third-party benchmark profile synced into LMSpeed; only metrics available for both models are shown.

Category performance

Compare benchmark category scores on a 0-100 scale. Select a category to inspect the gap.

Model A coverage
7 / 8
Model B coverage
7 / 8
Shared
7 shared categories

Avg. score

Claude Sonnet 4.6

52.7

Avg. score

GPT-5.4 Mini

50.9

Agents

Claude Sonnet 4.6 leads by 1.7

Claude Sonnet 4.651.9
GPT-5.4 Mini50.2

Coding

GPT-5.4 Mini leads by 0.5

Claude Sonnet 4.655.1
GPT-5.4 Mini55.6

Reasoning

GPT-5.4 Mini leads by 5.6

Claude Sonnet 4.649.6
GPT-5.4 Mini55.2

Knowledge

Claude Sonnet 4.6 leads by 25.2

Claude Sonnet 4.670.6
GPT-5.4 Mini45.4

Math

Claude Sonnet 4.6 leads by 3.3

Claude Sonnet 4.652.6
GPT-5.4 Mini49.3

Multilingual

No data

Claude Sonnet 4.6-
GPT-5.4 Mini-

Multimodal

GPT-5.4 Mini leads by 1.7

Claude Sonnet 4.645.4
GPT-5.4 Mini47.1

Instruction following

GPT-5.4 Mini leads by 9.9

Claude Sonnet 4.643.8
GPT-5.4 Mini53.7

Professional benchmark details

Metric-level scores with benchmark source, rank depth, confidence, error, and evaluation date where available.

Aggregatereported

BenchLM overall score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

59.0

Rank #53/111 · confidence 4 · eval date 2026-02-01

+1.0

GPT-5.4 Mini

winner

60.0

Rank #46/111 · confidence 4 · eval date 2026-03-17

Pricingverified

Input price

SourceClaude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

Claude Sonnet 4.6

$3.00/M

Rank #158/185 · confidence 4

+$2.25/M

GPT-5.4 Mini

winner

$0.750/M

Rank #105/185 · confidence 4

Pricingverified

Blended price

SourceClaude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

Claude Sonnet 4.6

$6.00/M

Rank #158/185 · confidence 4

+$4.31/M

GPT-5.4 Mini

winner

$1.69/M

Rank #112/185 · confidence 4

Pricingverified

Output price

SourceClaude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

Claude Sonnet 4.6

$15.00/M

Rank #158/185 · confidence 4

+$10.50/M

GPT-5.4 Mini

winner

$4.50/M

Rank #124/185 · confidence 4

Reasoningreported

BenchLM Reasoning score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

65.8

Rank #41/66 · confidence 4 · eval date 2026-02-01

+7.7

GPT-5.4 Mini

winner

73.5

Rank #30/66 · confidence 4 · eval date 2026-03-17

Reasoningreported

CritPt

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

0.9

Rank #75/88 · confidence 4 · eval date 2026-02-01

+9.1

GPT-5.4 Mini

winner

10.0

Rank #33/88 · confidence 4 · eval date 2026-03-17

Reasoningreported

AA-LCR

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

62.3

Rank #79/106 · confidence 4 · eval date 2026-02-01

+10.7

GPT-5.4 Mini

winner

73.0

Rank #45/106 · confidence 4 · eval date 2026-03-17

Reasoningverified

HLE

SourceClaude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

Claude Sonnet 4.6

winner

33.6%

Rank #48/214 · confidence 4

+15.0%

GPT-5.4 Mini

18.6%

Rank #89/214 · confidence 4

Reasoningverified

GPQA

SourceClaude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

Claude Sonnet 4.6

winner

87.5%

Rank #50/217 · confidence 4

+5.2%

GPT-5.4 Mini

82.3%

Rank #86/217 · confidence 4

Codingreported

Vibe Code Bench

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

51.5

Rank #7/33 · confidence 4 · eval date 2026-02-01

+3.5

GPT-5.4 Mini

48.0

Rank #10/33 · confidence 4 · eval date 2026-03-17

Codingreported

FrontierCode 1.1 Main

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

24.3

Rank #11/11 · confidence 4 · eval date 2026-02-01

+2.7

GPT-5.4 Mini

winner

27.0

Rank #9/11 · confidence 4 · eval date 2026-03-17

Codingreported

EEBench

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

20.4

Rank #14/25 · confidence 3 · eval date 2026-02-01

+2.1

GPT-5.4 Mini

18.3

Rank #16/25 · confidence 3 · eval date 2026-03-17

Codingreported

BenchLM Coding score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

62.0

Rank #22/86 · confidence 4 · eval date 2026-02-01

+5.8

GPT-5.4 Mini

56.2

Rank #34/86 · confidence 4 · eval date 2026-03-17

Codingreported

SWE-bench (Vals)

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

77.4

Rank #26/36 · confidence 4 · eval date 2026-02-01

+4.4

GPT-5.4 Mini

73.0

Rank #32/36 · confidence 4 · eval date 2026-03-17

Codingreported

LiveCodeBench (Vals)

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

82.1

Rank #29/34 · confidence 4 · eval date 2026-02-01

+0.6

GPT-5.4 Mini

81.5

Rank #30/34 · confidence 4 · eval date 2026-03-17

Codingreported

AA-SciCode

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

46.9

Rank #46/113 · confidence 4 · eval date 2026-02-01

+3.0

GPT-5.4 Mini

winner

49.9

Rank #33/113 · confidence 4 · eval date 2026-03-17

Mathreported

FrontierMath v2 (Tiers 1-3)

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

32.4

Rank #18/47 · confidence 4 · eval date 2026-02-01

+4.1

GPT-5.4 Mini

28.3

Rank #20/47 · confidence 4 · eval date 2026-03-17

Mathreported

FrontierMath v2 (Tier 4)

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

8.3

Rank #20/36 · confidence 4 · eval date 2026-02-01

+6.2

GPT-5.4 Mini

2.1

Rank #36/36 · confidence 4 · eval date 2026-03-17

Mathreported

BenchLM Math score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

49.0

Rank #38/63 · confidence 4 · eval date 2026-02-01

+4.5

GPT-5.4 Mini

44.5

Rank #39/63 · confidence 4 · eval date 2026-03-17

Knowledgereported

AA-Omniscience Hallucination Rate

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

68.5

Rank #57/106 · confidence 4 · eval date 2026-02-01

+21.7

GPT-5.4 Mini

winner

90.2

Rank #18/106 · confidence 4 · eval date 2026-03-17

Knowledgereported

MMLU-Pro (Vals)

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

87.3

Rank #22/37 · confidence 4 · eval date 2026-02-01

+2.7

GPT-5.4 Mini

84.6

Rank #31/37 · confidence 4 · eval date 2026-03-17

Knowledgereported

BenchLM Knowledge score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

75.3

Rank #26/82 · confidence 4 · eval date 2026-02-01

+18.3

GPT-5.4 Mini

57.0

Rank #61/82 · confidence 4 · eval date 2026-03-17

Knowledgereported

GPQA Diamond (Vals)

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

85.6

Rank #29/36 · confidence 4 · eval date 2026-02-01

+2.5

GPT-5.4 Mini

83.1

Rank #34/36 · confidence 4 · eval date 2026-03-17

Knowledgereported

AA-Omniscience Accuracy

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

38.6

Rank #40/106 · confidence 4 · eval date 2026-02-01

+1.1

GPT-5.4 Mini

37.5

Rank #44/106 · confidence 4 · eval date 2026-03-17

Knowledgereported

Artificial Analysis Intelligence Index

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

36.8

Rank #60/117 · confidence 4 · eval date 2026-02-01

+4.1

GPT-5.4 Mini

winner

40.9

Rank #46/117 · confidence 4 · eval date 2026-03-17

Knowledgereported

AA-GPQA Diamond

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

79.9

Rank #79/113 · confidence 4 · eval date 2026-02-01

+7.6

GPT-5.4 Mini

winner

87.5

Rank #47/113 · confidence 4 · eval date 2026-03-17

Knowledgereported

AA-HLE

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

13.3

Rank #82/114 · confidence 4 · eval date 2026-02-01

+14.8

GPT-5.4 Mini

winner

28.1

Rank #55/114 · confidence 4 · eval date 2026-03-17

Multimodalreported

BenchLM Multimodal Grounded score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

54.1

Rank #43/55 · confidence 4 · eval date 2026-02-01

+3.1

GPT-5.4 Mini

winner

57.2

Rank #40/55 · confidence 4 · eval date 2026-03-17

Multimodalreported

AA-MMMU-Pro

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

70.6

Rank #50/68 · confidence 4 · eval date 2026-02-01

+2.7

GPT-5.4 Mini

winner

73.3

Rank #43/68 · confidence 4 · eval date 2026-03-17

Instruction followingreported

BenchLM Instruction Following score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

47.9

Rank #49/51 · confidence 4 · eval date 2026-02-01

+41.7

GPT-5.4 Mini

winner

89.6

Rank #17/51 · confidence 4 · eval date 2026-03-17

Instruction followingreported

AA-IFBench

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

41.2

Rank #67/84 · confidence 4 · eval date 2026-02-01

+32.1

GPT-5.4 Mini

winner

73.3

Rank #24/84 · confidence 4 · eval date 2026-03-17

Agentsreported

OSWorld-Verified

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

72.1

Rank #16/27 · confidence 4 · eval date 2026-02-01

0.0

GPT-5.4 Mini

72.1

Rank #16/27 · confidence 4 · eval date 2026-03-17

Agentsreported

Terminal-Bench 2.1 (Vals)

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

winner

57.3

Rank #26/39 · confidence 4 · eval date 2026-02-01

+2.6

GPT-5.4 Mini

54.7

Rank #29/39 · confidence 4 · eval date 2026-03-17

Agentsreported

τ²-bench results

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

79.5

Rank #55/82 · confidence 4 · eval date 2026-02-01

+13.9

GPT-5.4 Mini

winner

93.4

Rank #29/82 · confidence 4 · eval date 2026-03-17

Agentsreported

Terminal-Bench 2.0

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

59.1

Rank #35/53 · confidence 4 · eval date 2026-02-01

+0.9

GPT-5.4 Mini

winner

60.0

Rank #31/53 · confidence 4 · eval date 2026-03-17

Agentsreported

BenchLM Agentic score

SourceClaude Sonnet 4.6

Claude Sonnet 4.6

50.5

Rank #54/76 · confidence 4 · eval date 2026-02-01

+4.8

GPT-5.4 Mini

winner

55.3

Rank #48/76 · confidence 4 · eval date 2026-03-17

API audit comparison

Latest completed audits from shared providers, with four safety and integrity score groups plus report links.

Provider
Claude Sonnet 4.6
GPT-5.4 Mini
No completed audits are available from shared providers yet.

Provider examples

Speed aggregates and input/output pricing share each provider row for real API selection and migration cost checks.

Provider
Claude Sonnet 4.6
GPT-5.4 Mini
UoCode25 tests

Claude Sonnet 4.6

speed / latency

32 tok/s / 6074ms

input / output

No data

GPT-5.4 Mini

speed / latency

N/A / N/A

input / output

No data

XShuLab Sub2API25 tests

Claude Sonnet 4.6

speed / latency

N/A / N/A

input / output

No data

GPT-5.4 Mini

speed / latency

139 tok/s / 2844ms

input / output

No data

天宫造物25 tests

Claude Sonnet 4.6

speed / latency

40 tok/s / 3145ms

input / output

No data

GPT-5.4 Mini

speed / latency

N/A / N/A

input / output

No data

Tokeness.io15 tests

Claude Sonnet 4.6

speed / latency

48 tok/s / 1587ms

input / output

No data

GPT-5.4 Mini

speed / latency

95 tok/s / 2343ms

input / output

No data

Mentoe API10 tests

Claude Sonnet 4.6

speed / latency

N/A / N/A

input / output

No data

GPT-5.4 Mini

speed / latency

181 tok/s / 3275ms

input / output

No data

Moyanjdc API

Claude Sonnet 4.6

claude-sonnet-4-6

speed / latency

No data

input / output

$0/M/$0/M

GPT-5.4 Mini

gpt-5.4-mini

speed / latency

No data

input / output

$0/M/$0/M

兔子API

Claude Sonnet 4.6

claude-sonnet-4-6

speed / latency

No data

input / output

$0/request

GPT-5.4 Mini

gpt-5.4-mini

speed / latency

No data

input / output

$0/request

Crond

Claude Sonnet 4.6

claude-sonnet-4-6-免费

speed / latency

No data

input / output

$0.0030/request

GPT-5.4 Mini

gpt-5.4-mini-免费

speed / latency

No data

input / output

$0.0030/request

钠 API

Claude Sonnet 4.6

[按次特价]claude-sonnet-4-6

speed / latency

No data

input / output

$0.0041/request

GPT-5.4 Mini

gpt-5.4-mini

speed / latency

No data

input / output

$0.011/M/$0.068/M

BUZZ

Claude Sonnet 4.6

claude-sonnet-4-6

speed / latency

No data

input / output

$0.035/M/$0.176/M

GPT-5.4 Mini

gpt-5.4-mini

speed / latency

No data

input / output

$0.0044/M/$0.026/M

FAQ

Weighted outcome: GPT-5.4 Mini. Benchmark capability categories carry 80%, while price, API performance, and availability carry 20%.

Why is this comparison indexable?
It has 6 verifiable comparison points, and both models have pricing or benchmark data.
Are missing metrics invented?
No. Metrics without LMSpeed data are omitted from this report.

Rankings are based on community-submitted tests and periodic health probes. Advisory only, not official data.